-
Notifications
You must be signed in to change notification settings - Fork 33
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Don't start Usage with Keras exampe #87
Comments
This makes sense to me. I'd start with this.
I don't think it's really about integrating with Keras though, or that we want to add specific integration pages (for now). That would be a separate question at least. So hopefully the Keras example can remain as an "advanced" section in the usage guide for now? (reused in the |
This comment has been minimized.
This comment has been minimized.
dvclive
: Update "Usage Guide"
I will start with that !
I think that even if it the integrations feel like a separate subject, the documentation should revolve around it. Reviewing other
I think that this makes a lot of sense given that most of the potential users of a "ml logger" will be probably already adepts of one of more "ml training" framework and the "low level" API of a "ml logger" doesn't really matters to them.
I usually see "advanced" sections as references for when the simplest solution doesn't cover my needs. However I feel like the current state of the keras example contains an unnecessary complexity that might mislead a potential user. We are telling the user to write a custom keras callback when there is one already implemented and maintained in the repository. In contrast, the DVC 2.0 release blog post is more concise about this and tells the user that a single import is needed. It would be useful to have an "advanced" section on how to write a custom integration, though. |
Hmmm maybe I'm confused about the term "integration" here. The only other software that DVCLive integrates with is DVC. Having usage examples with major frameworks does seem important too but again, let's make a separate issue for that? It shouldn't prevent us from keeping Keras in this main usage page (advanced section) IMO.
Right so I think we've already agreed to simplify the usage example at first. But are you saying we don't need an advanced section at all then? Just remove all Keras specific code and leave that for some other section about "integrations"? For now I wouldn't do that but it can be addressed by the other issue I'm recommending to be open. |
And advanced section would still make sense, I guess, but I'm not sure what should it cover. I see two types of use cases for Maybe the advanced section could dive into how to build new "Integrations"? Or "diving" into how some of the existing integrations are implemented (i.e. common
I think that we should go with that approach. An example of the
I think I will close this issue and try to create the task list you suggested in #86 |
The current
dvclive
Usage Guide directly jumps into thekeras
integration omitting the possibility of it's most basic usage (as a python library). Likewise theDvclive with DVC
is also "tied" to thiskeras
example.I think that it might be better to start the Usage Guide with a standalone approach (something similar to this existing section of the README) and a separate page explaining the usage along with DVC and link to a new page called
Integrations
.The current content of the Usage Guide could be moved to a new
Integrations/keras
section. And we might create new pages for other undocumented integrations likexgboost
ormmcv
. Further integrations could be simply appended toIntegrations
.Something like:
The text was updated successfully, but these errors were encountered: